New technologies in groundwater exploration. Surface ......Geologica Acta, Vol.2, Nº2, 2004, 109-120 U. YARAMANCI The new SNMR method in groundwater exploration 111 where is the gyromagnetic

New technologies in groundwater exploration.Surface Nuclear Magnetic Resonance

As groundwater becomes increasingly important for living and environment, techniques are asked for animproved exploration. The demand is not only to detect new groundwater resources but also to protect them.Geophysical techniques are the key to find groundwater. Combination of geophysical measurements with bore-holes and borehole measurements help to describe groundwater systems and their dynamics. There are a num-ber of geophysical techniques based on the principles of geoelectrics, electromagnetics, seismics, gravity andmagnetics, which are used in exploration of geological structures in particular for the purpose of discoveringgeoresources. The special geological setting of groundwater systems, i.e. structure and material, makes it neces-sary to adopt and modify existing geophysical techniques. A new discipline called hydrogeophysics has beenformed and is growing fast. Efforts for direct detection of groundwater led recently to a new technique: SurfaceNuclear Magnetic Resonance (SNMR). The principle of nuclear magnetic resonance, well known in physics,physical chemistry as well as in medicine, has successfully been adapted to assess the existence of groundwaterand the aquifer parameters. This technique allows for the first time detecting and assessing water directly byonly surface measurements allowing quantitative information about mobile water content as well as pore struc-ture parameters leading to hydraulic conductivities. Function, results, interpretation, advantages and drawbacksof SNMR are reviewed in this paper showing the current state of art and developments. A comprehensive exam-ple of SNMR is presented with measurements conducted at the site of Nauen near Berlin. The site has Quater-nary aquifers with differing layering of sand and till. The results are very satisfying as aquifers down to 50 mdepth can be identified quite reliably. The water content is estimated with a high degree of accuracy and relax-ation times allowed to derive hydraulic conductivities. Supplementary measurements with geoelectrics andradar made possible to complement and confirm the information achieved with SNMR as well as to apply ajoint multimethod approach to aquifer assessment.

Ground water exploration. Surface Nuclear Magnetic Resonance. Water content. Pore size.

INTRODUCTION

Hydrogeological systems and hydrological processesare quite complex. There are still some phenomena whichare not well understood yet and not even properly

described or observed. As for the study of any complexsystem there is a great need and demand of improving thetechnology for the investigations. Very often the improve-ments are on measuring and processing of currently usedmethods. This includes collection of more data in a more

Geologica Acta, Vol .2 , Nº2, 2004, 109-120

Avai lable onl ine at www.geologica-acta.com

© UB-ICTJA 109

KEYWORDS

A B S T R A C T

U. YARAMANCI

Technical University of Berlin, Department of Applied GeophysicsAckerstr.71-76, 13355 Berlin, Germany. E-mail: [email protected]

Geolog ica Acta , Vo l .2 , Nº2, 2004, 109-120 110

The new SNMR method in groundwater exploration U. YARAMANCI

accurate, faster and cheaper way. Along with theimprovements of computers in speed and storage alsoprocessing of data gets faster and in particular new algo-rithms can be applied. These kinds of improvements areresult of continuous efforts on developing technology inhardware and software. Besides these improvements thereis a need on a better access to some properties ofhydrosystems, which allow an improved understanding,description and prediction of the behaviour of these sys-tems.

In comparison to the almost continuous improvementsof existing technology, very rarely in geophysics a com-pletely new technology or approach will emerge which isdifferent from the ones already existing. In the followingone such new technology will be presented which justpassed the experimental stage to become a promising andvaluable complementary tool for investigating hydrosys-tems. Surface Nuclear Magnetic Resonance (SNMR)allows for the first time detecting and assessing waterdirectly by only surface measurements allowing quantita-tive information of mobile water content and pore struc-ture parameters leading to hydraulic conductivities. Thetechnique is occasionally called also Magnetic ResonanceSounding (MRS) or Proton Magnetic Resonance (PMR).

The SNMR method is a fairly new technique in geo-physics to investigate directly the existence, amount andproducability of groundwater by measurements at the sur-face. The first high-precision observations of nuclearmagnetic resonance (NMR) signals from hydrogen nucleiwere made in the forties. This technique has found wideapplication in chemistry, physics, tomographic imaging inmedicine, as well as in geophysics. It is also a standardinvestigation technology on rock cores and in boreholes(Kenyon, 1992). The first ideas for making use of NMRin groundwater exploration from the ground surface weredeveloped as early as the 1960s, but only in the 1980swas effective equipment designed and put to operation forsurface geophysical exploration (Semenov et al., 1988;Legchenko et al., 1990). Extensive surveys and testinghave been conducted in different geological conditionsparticularly in sandy aquifers but also in clayey forma-tions as well as in fractured limestone and at special testsites (Schirov et al., 1991; Lieblich et al., 1994; Goldmanet al., 1994; Legchenko et al., 1995; Beauce et al., 1996;Yaramanci et al., 1999; Meju et al., 2002; Plata and Rubio,2002; Supper et al., 2002; Vouillamoz et al., 2002; Yara-manci et al., 2002) revealing the power of the method aswell as the shortcomings which need to be improved.

BASIC PRICIPLES OF SNMR

The protons of the hydrogen atoms in water moleculeshave a magnetic moment . They can be described in

terms of a spinning charged particle. Generally isaligned with the local magnetic field B0 of the Earth.When another magnetic field B1 is applied, the axes ofthe spinning protons are deflected, owing to the torqueapplied (Fig. 1). Hereby only the component of B1 per-pendicular to the static field B0 acts as the torque force.When B1 is removed, the protons generate a relaxationmagnetic field as they become realigned along B0 whileprecessing around B0 with the Larmor frequency

fL = L / 2 with L = B0 and = 0.267518 Hz/nT (1)

FIGURE 1 (A) Elements of SNMR measurements and (B) behaviorof relaxing protons.

A

B

Geolog ica Acta , Vo l .2 , N º2, 2004, 109-120


111

where is the gyromagnetic ratio of hydrogen protons.The measurements are conducted using a loop usuallywith a circular or rectangular lay out. An alternating cur-rent,

i(t) = i0 cos(L t), (2)

with the Larmor frequency L and strength i0 is passedthrough the loop for a limited time so that an excitationintensity (pulse moment) of q = i0 is achieved. After thecurrent in the loop is switched off, a voltage e(t) with thefrequency L and decaying amplitude is induced in theloop by the relaxation of the protons:

e(t) = L M0 f(r) e-t/T(r) cos(L t + (r))B (r) sin(0.5 B (r) q) dV. (3)

Here is M0 = 3.2910-3 B0 J/(T m3), the nuclear magneti-sation for water at a temperature of 293 K. In a unit vol-ume dV at the location r(x,y,z) the volume fraction ofwater is given by f(r) and the decay time of protons byT(r). B (r) is the component of the exciting field (nor-malised to 1 A) perpendicular to the static magnetic fieldB0 of the Earth. In a conductive medium B (r) is com-posed of the primary field of the loop and the induced field.The induced field causes a phase shift (r) in respect to theexciting field. The argument of the sine function, = 0.5 B (r) q, is the angle of deflection of the magnetic momentof the protons from the magnetic field of the Earth. Thesignal e(t) is usually approximated by (Fig. 2)

e(t) = E0 e-t/T cos(L t + ). (4)

The envelope of this voltage is directly related to thewater content and to the decay time of every volume ele-ment in the underground contributing to the signal: Fornon conductive media, i.e. negligible phases, the initialamplitude E0 at t=0 is related only to the water content:

E0 = 0 M0 f(r) B (r) sin(0.5 B (r) q) dV. (5)

Using this equation the initial amplitudes for variousexcitations intensities and for water layers at differentdepths and thicknesses can be calculated. It happens thatfor deeper water layers the maximum of E0 occurs athigher q values and the strength of E0 is directly related tothe thickness of the layer and the amount of the water.

The recorded decay time T is the relaxation-time con-stant (spin-spin or transversal relaxation time) denotedusually with T2

* in the usual NMR terminology. It isrelated to the mean pore size and, therefore, grain size aswell as hydraulic conductivity of the material. Clay,including sandy clays, usually has a decay time of lessthan 30 ms, whereas sand has one of 60 - 300 ms, gravel

300 - 600 ms, and pure water 600 - 1000 ms (Schirov etal., 1991).

The measurements are conducted for different excita-tion intensities q and the main parameters recorded forevery q are the initial amplitude E0 and the decay time T.This set of data in form of E0(q) and T(q) are inverted tofind the distribution of water content with depth f(z) andof decay time with depth T(z). Thereby equation (3) is thebasis of the inversion which is used in the form modifiedfor models having horizontal layering. Two or threedimensional inversion and suitable measurementapproaches for this are not available yet but are currentlybeing investigated. The usual inversion scheme used inSNMR is based on a least square solution with a regulari-sation (Legchenko and Shushakov, 1998). Lately newinversion schemes are developed which use model opti-misation with Simulated Annealing and allow more flexi-bility in designing the layer thicknesses imposed oninversion even with free layer thicknesses to be optimised(Mohnke and Yaramanci, 1999, 2002).

INVESTIGATIONS AT THE SITE NAUEN

To demonstrate the field work, advantages and draw-backs of SNMR a case history from the test site in Nauenis very suitable and will be briefly presented here extract-ed from a previous detailed work (Yaramanci et al.,2002). The geology of the Nauen site is quite typical forlarge areas in Northern Germany. It is build up of Quater-nary sediments emerged in the last glacial period (Weich-sel) and overlying Tertiary clays. In the actual area ofinvestigation these sediments mainly consist of fluvial

FIGURE 2 Input and output signal in the SNMR measurements.

Geolog ica Acta , Vo l .2 , N º2, 2004, 109-120 112


sands bordered by glacial till. The topography is charac-terised by flat hills underlain by till and plains comprisingglacial fluvial sands and gravels. In the test area there isan unconfined shallow aquifer consisting of fine to medi-um sands underlain by an aquiclude of marly and clayeyglacial till. North of the site the glacial till approaches thesurface in a nearly E-W strike. For this reason all themain profiles were chosen N-S.

SNMR measurements

The SNMR soundings were measured at 5 locations,25 m apart from each other on the main profile usingeight shaped loop antennas with 50 m in diameter. Theantenna main axis was directed E-W, parallel to thestrike. The simple circular or square loop, as used for anearlier single sounding, could not be used because of thehigh noise, generated by an electrical railway some 5km away. The eight shaped loop allowed to increase thesignal to noise ratio very effectively, almost ten timescompared with simple loops. Therefore, moderate stack-ing rates of 32 were quite sufficient; 16 to 24 differentexcitation intensities (q) levels were used. The frequen-cies have been around 2082 Hz, corresponding to thetotal intensity of local Earth magnetic field about 48900nT, very stable within ± 0.5 Hz for all excitation intensi-ties. Measurements were conducted with the NUMISsystem (NUMIS, 1996). The processing and inversion ofthe data were made with the standard inversion softwareprovided with this system (Legchenko and Shushakov,1998).

In Fig. 3 an example of a raw data set is shown forthe location B8. These data document the good qualityof the measurements and is representative for the othersoundings as well. The decaying is well recognisablebecause the amplitudes of stacked noise are about 25–50 nV, which is quite low compared to the amplitudeof the SNMR signal, in particular to the initial ampli-tude.

In Fig. 4B the actual measurements of amplitudes,decay times, phases and stacked noise for different excita-tion intensities are shown. The amplitudes, showing aprominent maximum of about 550–750 nT for q 400 A ms, qualitatively suggest that a significantaquifer is present at a very shallow depth. Decay timesare around 100–200 ms, suggesting sandy material. Allphases start at 0o, increasing up to about 60o with increas-ing excitation intensity. This usually points to an increas-ing conductivity with depth.

In Fig. 4A the results of inversion are shown. Theinversion scheme used is a least squares type of algorithmusing a regularisation parameter, affecting the smoothnessof water content distribution with depth (Legchenko and

Shushakov, 1998). The range of smoothness might bevery high leading to different interpretation of the watercontent distributions (Yaramanci et al., 1998). In particu-lar by smooth inversion it is not possible to see sharpboundaries or changes of water content with depth as itwould be expected for aquifers in medium to coarse sandswith small capillary fringes or in the presence of a distinctaquiclude as it is the case here. Therefore a moderate reg-ularisation was used which does not smooth the structuresvery drastically. A low regularisation would smooth lessbut may cause unrealistic variations for small depthranges which are not conform with the resolution abilityof the method.

The water contents are about 20% near surface,increasing up to 30% at depths of 10 m followed by astrong decrease down to 5% in a depth range of around25 m. In deeper areas the water contents get unusuallyhigh; even up to 40%. The upper part of the picture,showing clearly an aquifer, is in quite good agreementwith the expected and partly known geology. The lowwater content range must correspond to the glacial tillbecause of the very low amount of mobile water in thattype of geological material. The increase of water contentbelow the till suggests a second aquifer which is con-firmed with a recent borehole.

The inverted decay times are in the range between100 - 300 ms. The common feature of all decay timedistributions is the maximum of roughly 200 ms atabout 20 m and an average value of 150 ms at a depthof about 10 m, thus corresponding to the maximumwater content. Otherwise there is no general similarityi.e. decay times increase or decrease to greater andshallower depths. The range in general shows fine tomedium sands, even coarse sands. The higher decaytimes at the depth of low water content is difficult toexplain for the geological conditions at this site. Thiswould be only possible for low porosity material butwith relative high degree of mobile water content i.e.compact sediments with little cementation or fracturedhard rock, but not for glacial till. It is not clear yet,whether this problem is due to the use of the specialeight shaped loop antenna and/or the correspondinginversion as the signal is small from this depth. In anearlier measurement with square loop the inverteddecay times show the expected behaviour (Yaramanciet al., 1998) whereas the water contents are the same asthose with eight shaped loop.

Radar measurements

An extensive GPR survey has been carried out at thetest site. For the measurements 200 MHz antennas wereused to achieve an optimum in penetration and resolution.The profiles were orientated in N-S and E-W directions

Geolog ica Acta , Vo l .2 , N º2, 2004, 109-120


113

FIGURE 3 Example of a SNMR data set in the location B8 at the test site Nauen. Time axes are in ms. Envelopes of in-phase and out-phase compo-nents, X(t) and Y(t), are measured to yield the amplitude E(t) = (X(t)2+Y(t)2)1/2 and phase (t) = arctan (Y(t)/X(t)). The parameters used for the inver-sion are E0 = E(t=0) and the decay time T, determined by an exponential fitting of e-t/T to E(t). The actual phase of the signal is 0 = (t=0).

consisting of 4 sections each 25 m apart. An example of aN-S sections, perpendicular to geological strike, is shownin Fig. 5. Also shown is the CMP (Common-mid-point)measurement at this profile in order to derive propervelocities for time to depth conversions and rock physicalconsiderations.

GPR can be used to map water table by radar reflec-tion if there is a sharp discontinuity and not a large transi-tion zone due to the capillary fringe. Usually the aquiferwill not be penetrated properly to give reflections from itsbottom due to the high absorption, i.e. energy loss in theaquifer. In Nauen both favourable situations are met, thatthere is a sharp water table due to the high hydraulic con-ductivity of the sand, and therefore a good reflection andthere is low electrical conductivity of water and thereforegood bottom reflections, i.e. reflections from the top ofglacial till.

According to GPR measurements the water table is atabout 2–3 m depth depending on the slowly varyingtopography. The bottom of the aquifer, i.e. the top ofglacial till, is about 15 m at the south and graduallycomes up on the surface in northern direction. North ofthis, below the top of till, there is no radar signal due tohigh absorption.

Geoelectric measurements

The clear indication of a 2-dimensional structure inGPR measurements was the reason for performing 2D-geoelectric measurement in order to get more detailedinformation about the resistivity structures.

Different geoelectric sections were measured withvarious lengths and electrode spacings. The result of themeasurement at the same profile as for radar measure-ments, where a Wenner array with an electrode spacing of2 m was used, is shown in Fig. 5. The result of a standardsmooth inversion (Loke and Barker, 1996) shows wellrecognisable structures of high resistivities at shallowranges in the south, corresponding to the vadose zone,and medium resistivities below, corresponding to theaquifer and glacial till. To the north this low resistivitylayers come up to the surface and the values of resistivityget even lower.

The meanwhile classical inversion of geoelectricpseudosections, imposing a smoothness on the resistivi-ties, has some drawbacks in case of well defined struc-tures with sharp resistivity contrasts (Olayinka and Yara-manci, 2000). Not only the boundaries of the structuresare blurred, but also the resistivities are far from true val-ues. In cases like Nauen, where sharp boundaries occur,block inversion may be used (Fig. 5). By this kind ofinversion distinct resistivities are found for individualblocks i.e. formations. The resistivities found are:3000 m for the vadose zone, 280 m for the aquiferand 90 m for the glacial till. These values are in verygood agreement with those known from other geoelectri-cal surveys at Quaternary aquifers in the Berlin region.The unusually low resistivity of 34 m for shallowranges within the glacial till in the northern part of the

Geolog ica Acta , Vo l .2 , N º2, 2004, 109-120 114


FIGURE 4 SNMR measurements and results at the test site Nauen.

A

B


115Geolog ica Acta , Vo l .2 , N º2, 2004, 109-120

FIGURE 5 Geoelectric and radar measurements at the test site Nauen at the same profile as for SNMR measurements.

profile shows an internal structure, probably reflectinghigher clay content.

ESTIMATION OF WATER CONTENT

The water content for the aquifer indicated bySNMR is about 25–30% (Fig. 4). That corresponds tothe mobile (free) water in the pores. This is an averagevalue for at least 3 locations to the south, B8, B7 andB2. Although the sharp boundary of water table is notwell determined by the SNMR inversion used, thevadose zone with a mean water content of 10–20 % isdistinguishable to some degree here. This water corre-sponds to the seeping water which needs some time toreach the aquifer. Below the aquifer a material withsome 5–10 % water content is shown, which should bethe glacial till. This, in fact, is in agreement with freewater contents which can still be accommodated inglacial till. The higher water contents found for deeperregions, suggesting a second aquifer, are confirmedwith a recent borehole.

Estimation of water content from radar measure-ments

The estimation of water content with GPR is quitestraightforward and widely used (Greaves et al., 1996;Hubbard et al., 1997; Dannowski and Yaramanci, 1999).The dielectric permittivity of the rock can be determinedby the velocity of electromagnetic waves measured withradar using

= (c/v)2 (6)

with c the velocity of electromagnetic waves in air. Toestimate water content from radar data the usual CRIMrelation (Mavko et al., 1998) has been used. According tothis formula the dielectric permittivity of the rock is relat-ed to the properties of the rock components by

M = (1 )Mm S Mw (1 S) Ma (7)

with dielectric permittivity of rock, porosity, S degreeof saturation, w dielectric permittivity of water = 80, m

dielectric permittivity of matrix 4.5, a dielectric per-mittivity of air = 1. The porosity is defined as = Vp/Vand the degree of saturation as S = Vw/Vp, with V, Vpand Vw being the volumes of the rock, pores and waterrespectively. The water content G = Vw/V is then givenby

G = S (8)

In case of (full) saturation i.e. S = 1 the relation (7) turnsto be (9)

M = (1 )Mm Mw (9)

and water content is equal to porosity, G = . The porosi-ty determined using (9) is then used in (7), to estimate thewater content in the vadose zone with the dielectric per-mittivity of the vadose zone.

Estimation of water content from geoelectricmeasurements

In order to interpret the resistivity and its local varia-tions, the physical cause of resistivity and the influencingfactors must be well understood. The general model ofrock conductivity is described by two conductivities ina parallel circuit and therefore in addition (Gueguen andPalciauskas, 1994; Mavko et al., 1998)

= v + q (10)

v is the volume conductivity caused by the ionicconductivity of the free electrolyte in the pores and q thecapacitive interlayer conductivity due to adsorbed waterat the internal surface of the pores. The conductivity qis, in contrast to v, strongly frequency dependent - beingvery small for zero frequency and becoming large withincreasing frequency. For rocks with a large internal sur-face, for example containing a great portion of clay, qmight be very high and it is therefore called the “clayterm”. For aquifers in more sandy formations and at lowfrequencies v is much larger than q, so that v.Going back to the more familiar expression in terms ofresistivity, with = 1/, the ohmic resistivity of rock is

= w -m

S-n

= w F I (11)

where w is the resistivity of water, the porosity, m theArchie exponent (or cementation factor), S the degree ofsaturation, n saturation exponent. The actual dependenceof the resistivity on the pores is expressed with the forma-tion factor F = -m and saturation index I = S-n. For a sat-urated rock (i.e. S=1) the resistivity in (11) becomes

o = w -m

(12)

where the index o stands for (fully) saturated rock. This isthe well known Archie equation (1942), which is widelyused, particularly in interpretation of resistivity well logs.

Equation (12) is used first to estimate the water con-tent in the aquifer and the porosity determined hereby isthen used for the estimation of porosity in the vadosezone. Very often it can be assumed that m - n 0 andconsequently Sm-n 1. Even though the effect of S is notthat high, it should not be neglected a priori, particularlyin such cases, when values for m and n are available. The



parameters used for the estimation of porosity and watercontent at the test site Nauen and the correspondingresults are shown in Table 1.

ESTIMATION OF HYDRAULIC CONDUCTIVITIESFROM SNMR

In order to obtain hydraulic conductivities fromSNMR measurements, the empirical relationshipbetween decay time and average grain size observed inmany SNMR surveys (Schirov et al., 1991) can beused. Combining this with the relationship betweengrain size and hydraulic conductivity often used inhydrogeology (Hölting, 1992) leads to a simple estima-tion of hydraulic conductivity as proposed by Yara-manci et al. (1999):

kf T4 (13)

with kf the hydraulic conductivity in m/s and T the decaytimes in s. An average decay time of 150 ms yields ahydraulic conductivity of 5x10-4 m/s which fits quite wellwith the hydraulic conductivities determined on corematerial.

DISCUSSION

In the SNMR measurements at Nauen site an aquifercan easily be identified at a shallow depth range, followedby a low water content layer. The boundaries are not verywell defined, as the used standard NUMIS inversionscheme favours and even imposes a smooth distributionof water content with depth. Thus, the small vadose zoneis not well resolved, but still detected. In contrast GPRallowed a very good determination of the structure withclear reflections from the water table and the top ofglacial till. This situation is, however, not usual that waywith GPR elsewhere. The resistivity distribution derivedfrom 2D-geolelectrical measurements shows a generaltrend, indicating the vadose zone and the aquifer and eventhe glacial till below. But the obtained results are toosmooth to allow the determination of exact depths of lay-ers. The block inversion here yields more reliable trueresistivities of the layers but needs some a priori informa-tion about the geometry.

The estimation of water content with SNMR workedvery well showing 30% mobile water in the aquifer whichis in extraordinary good agreement with the independentmeasurements in the laboratory. The water content shownby SNMR for the vadose zone is somewhat higher as its



RADAR GEOELECTRICS SNMR

sounding 2D-section

Vadose zone 5% 4% 5% 15-20%

0 – 2 musing: using: using:

v = 0.130 m/ns ρ0 = 8x103 Ωm ρ0 = 3x103 Ωmε = 5.3 φ = 28% φ = 20%

φ = 20% ρw = 33 Ωm ρw = 33 Ωmεw = 80 m = 1.3 m = 1.3εm = 4.5

εa = 1

Aquifer 20% 28% 20% 25-30%2 – 13 m

using: using: using:v = 0.86 m/ns ρ0 = 170 Ωm ρ0 = 280 Ωm

ε = 12.2 ρw = 33 Ωm ρw = 33 Ωmεw = 80 m = 1.3 m = 1.3εm = 4.5

TABLE 1 Estimation of water contents (in Vol. %) with different methods at the test site Nauen/Berlin(1)

(1) The estimations are first conducted for the aquifer. Derived water contents in the aquifer equal to the porosities which are then used forthe estimations in the vadose zone. Take note of that the SNMR estimations relate to free water and estimations from GPR and Geoelectricsrelate to total water.

true value can be, but this is caused by the problems toresolve the thin vadose zone. Even under very favourableconditions for interpreting 2D-geoelectrics - the structureis known from GPR and used in geoelectric inversion,thus producing more reliable resistivities - the estimatedwater contents are far from reality. The seemingly goodresult from geoelectric sounding is not real as thereshould be a large influence by side effects.

CONCLUSIONS AND DEVELOPMENTS

The method of SNMR has just passed the experimen-tal stage to become a powerful tool for groundwaterexploration and aquifer characterisation. Some furtherimprovements are still necessary and just in work. Thelargest concern currently is the effect of resistivities andtheir inclusion into the analysis and inversion as the excit-ing field will be modified and polarised considerably bythe presence of conductive structures. Earlier considera-tions of this (Shushakov and Legchenko, 1992;Shushakov, 1996) have led to appropriate theoreticaldescription and numerical handling of this problem (Vallaand Legchenko, 2002; Weichmann et al., 2002). In factthe incorporation of resistivities allows modelling of thephases in a reliable way (Braun et al., 2002) which is notonly useful for understanding the phases measured butalso the basis for a successful inversion of phases to yieldthe resistivity information directly from SNMR measure-ments.

In the analysis of SNMR the relaxation is generallyassumed to be monoexponential. Even if individual layerswere monoexponential in decaying, the integration resultsin a multiexponential decay in the measured signal. Themost comprehensive way taking account for this is toconsider decay time spectra in the data as well as in theinversion (Mohnke et al., 2001). This leads to the poresize distribution which is a new information and allowsimproved estimation of hydraulic conductivities. The esti-mation of water content also will be improved significant-ly as the initial amplitudes are much better determinedusing decay spectra approach.

Currently SNMR is carried out with a 1-D workingscheme. However, the errors might be very large byneglecting the 2-D or even 3-D geometry of the structures(Warsa et al., 2002) which have to be considered in theanalysis and inversion in the future. Measurement layouts are to be modified to meet the multi-D conditions,which is easier to accomplish for nonconductive struc-tures. In multi-D structures the actual difficulty is thenumerical incorporation of the electromagnetic modellingfor the exciting field.

As in any geophysical measurement, also with SNMR

the inversion plays a key role by interpreting the data.The limits of inversion and also the imposed conditions interms of geometrical boundary conditions as well as dif-ferences in the basic physical model used may lead toconsiderable differences. The inversion of SNMR datamay be ambiguous, since not only different regularisa-tions in the inversion impose a certain degree of smooth-ness upon the distribution of water content (Legchenkoand Shushakov, 1998; Yaramanci et al., 1998; Mohnkeand Yaramanci, 1999) but also the number of layers andthe size of layers forced in the inversion may consider-ably effect the results. The rms-error is not necessarily asufficient measure for assessing the quality of the fit of amodel to the observed data. The most recent research sug-gests that a layer modelling with free boundaries avoidsthe problems associated with regularisation and takes intoaccount the blocky character of the structure whereappropriate (Mohnke and Yaramanci, 2002).

Further improvement in the inversion can be achievedif geoelectrical measurements are available and they canbe incorporated into a joint inversion with SNMR. Exam-ples of joint inversion of SNMR with Vertical ElectricalSounding show considerable improvement in thedetectability and geometry of the aquifers and allowsalso, by utilising of appropriate petrophysical models, theseparation of mobile and adhesive water (Hertrich andYaramanci, 2002).

At sites where no information is available in advance,SNMR should always be carried out along with geoelec-trical methods, i.e. direct current geoelectric, electromag-netics and even GPR. This will help to decrease ambigui-ty in the results and also allow hydrogeologicalparameters to be estimated (Yaramanci et al., 2002). Butdespite all the difficulties, the quality of geophysicalexploration for groundwater and aquifer properties willhave an increased degree of reliability by using SNMR asa direct indicator of water and soil properties.

The importance of the SNMR method lies in its abilityto detect water directly and allowing reliable estimationof mobile water content and hydraulic conductivity. Inthis respect it is unique, since all other geophysical meth-ods yield estimates, if ever, indirectly via resistivity,induced polarisation, dielectric permittivity or seismicvelocity. Using SNMR in combination with other geo-physical methods not only allows direct assessment but iscomplementary to the information yielded by other geo-physical methods.

REFERENCES

Archie, G.E., 1942. The electrical resistivity as an aid in deter-mining some reservoir characteristics. Transactions of



Americican Institute of Mineralogy, Metalogy and Petrole-um Engineers, 146, 54-62.

Beauce, A., Bernard, J., Legchenko, A., Valla, P., 1996. Unenouvelle méthode géophysique pour les étudeshydrogéologiques: l’application de la résonance magnétiquenucléaire. Hydrogéologie, 1, 71-77.

Braun, M., Hertrich, M., Yaramanci, U., 2002. Modeling ofphases in Surface NMR. Proceedings of 8th European Meet-ing on Environmental and Engineering Geophysics, Aveiro,Portugal, ed. EEGS-ES, 253-256.

Dannowski, G., Yaramanci, U., 1999. Estimation of water con-tent and porosity using combined radar and geoelectricalmeasurements. European Journal of Environmental andEngineering Geophysics, 4, 71-85.

Goldman, M., Rabinovich, B., Rabinovich, M., Gilad, D., Gev,I., Schirov, M., 1994. Application of the integrated NMR-TDEM method in groundwater exploration in Israel. Journalof Applied Geophysics, 31, 27-52.

Greaves, R.J., Lesmes, D.P., Lee, J.M. Toksöz, M.N. 1996.Velocity variations and water content estimated from multi-offset, ground penetrating radar. Geophysics, 61, 683-695.

Gueguen, Y., Paliciauskas, V., 1994. Physics of Rocks. Prince-ton, Princeton University Press, 294 pp.

Hertrich, M., Yaramanci, U., 2002. Joint inversion of SurfaceNuclear Magnetic Resonance and Vertical Electrical Sound-ing. Journal of Applied Geophysics, 50, 179-191.

Hölting, B., 1992. Hydrogeologie. Stuttgart, ed. Enke Verlag,418 pp.

Hubbard, S.S., Peterson, J.E., Jr., Majer, E.L., Zawislanski, P.T.,Wiliams, K.H., Roberts, J., Wobber, F., 1997. Estimation ofpermeable pathways and water content using tomographicradar data. The Leading Edge, 16, 1623-1628.

Kenyon, W.E., 1992. Nuclear magnetic resonance as a petro-physical measurement. International Journal of Radiation,Applications Instrumentation, Part E, Nuclear Geophysics,6, 153-171.

Legchenko, A.V., Semenov, A.G., Schirov, M.D., 1990. Adevice for measurement of subsurface water saturated layersparameters (in Russian). USSR Patent 1540515, 12 pp.

Legchenko, A.V., Shushakov, O.A., 1998. Inversion of surfaceNMR data. Geophysics, 63, 75-84.

Legchenko, A.V., Shushakov, O.A., Perrin, J.A., Portselan,A.A., 1995. Noninvasive NMR study of subsurface aquifersin France. Proceedings of 65th Annual Meeting of Society ofExploration Geophysicists, Houston, USA, 365-367.

Lieblich, D.A., Legchenko, A., Haeni, F.P., Portselan, A.A.,1994. Surface nuclear magnetic resonance experiments todetect subsurface water at Haddam Meadows, Connecticut.Proceedings of the Symposium on the Application of Geo-physics to Engineering and Environmental Problems,Boston, 2, 717-736.

Loke, M.H., Barker, R.D., 1996. Rapid least-squares inversionof apparent resistivity pseudosections by a quasi-Newtonmethod. Geophysical Prospecting, 44, 131-152.

Mavko, G., Mukreji, T., Dvorkin, J., 1998. The Rock PhysicsHandbook. Cambridge, Cambridge University Press, 330 pp.

Meju, M.A., Denton, P., Fenning, P., 2002. Surface NMRsounding and inversion to detect groundwater in keyaquifers in England: comparisons with VES–TEM methods.Journal of Applied Geophysics, 50, 95-111.

Mohnke, O., Yaramanci, U., 1999. A new inversion scheme forsurface NMR amplitudes using simulated annealing. Pro-ceedings of 60th Conference of European Association ofGeoscientists and Engineers, Helsinki, Finland, ed. EAGE,2-27.

Mohnke, O., Yaramanci, U., 2002. Smooth and block inversionof surface NMR amplitudes and decay times using simulat-ed annealing. Journal of Applied Geophysics, 50, 163-177.

Mohnke, O., Braun, M., Yaramanci, U., 2001. Inversion ofdecay time spectra from Surface-NMR data. Proceedingsof 7th European Meeting on Environmental and Engineer-ing Geophysics, Birminghan, Great Britain, ed. EEGS-ES.

NUMIS, 1996. Manual of NUMIS system for Proton MagneticResonance, IRIS Instruments, 56 pp.

Olayinka, A.I., Yaramanci, U., 2000. Use of block inversion inthe 2-D interpretation of apparent resistivity data and itscomparison with smooth inversion. Journal of Applied Geo-physics, 45, 63-81.

Plata, J., Rubio, F., 2002. MRS experiments in a noisy area ofdetrital aquifer in the south of Spain. Journal of AppliedGeophysics, 50, 83-94.

Schirov, M., Legchenko, A., Creer, G., 1991. A new direct non-invasive groundwater detection technology for Australia.Exploration Geophysics, 22, 333-338.

Semenov, A.G., Burshtein, A.I., Pusep, A.Y., Schirov, M.D.,1988. A device for measurement of underground mineralparameters (in Russian). USSR Patent 1079063.

Shushakov, O.A., 1996. Groundwater NMR in conductivewater. Geophysics, 61, 998-1006.

Shushakov, O.A., Legchenko A.V., 1992. Calculation of the pro-ton magnetic resonance signal from groundwater consider-ing the electroconductivity of the medium (in Russian).Novosibirsk, Russian Academy of Science, Institute ofChemical Kinetics and Combustion, 36, 1-26.

Supper, R., Jochum, B., Hübl, G., Römer, A., Arndt, R., 2002.SNMR test measurements in Austria. Journal of AppliedGeophysics, 50, 113-121.

Valla, P., Legchenko, A., 2002. One-dimensional modelling forproton magnetic resonance sounding measurements over anelectrically conductive medium. Journal of Applied Geo-physics, 50, 217-229.

Vouillamoz, J.-M., Descloitres, M., Bernard, J., Fourcassier, P.,Romagny, L., 2002. Application of integrated magnetic res-onance sounding and resistivity methods for borehole imple-mentation. A case study in Cambodia. Journal of AppliedGeophysics, 50, 67-81.

Warsa, W., Mohnke, O., Yaramanci, U., 2002. 3-D modelling ofSurface-NMR amplitudes and decay times. Third Interna-tional Conference on Water Resources and Environment,Dresden, July 2002, Eigenverlag des Forums für Abfall-wirtschaft und Altlasten e.V., 209-212.



Weichman, P.B., Lun, D.R., Ritzwoller, M.H., Lavely, E.M.,2002. Study of surface nuclear magnetic resonance inverseproblems. Journal of Applied Geophysics, 50, 129-147.

Yaramanci, U., Lange, G., Hertrich, M., 2002. Aquifer charac-terisation using Surface NMR jointly with other geophysicaltechniques at the Nauen/Berlin test site. Journal of AppliedGeophysics, 50, 47-65.

Yaramanci, U., Lange G., Knödel K., 1998. Effects of regularisa-tion in the inversion of Surface NMR measurements. 60th

Conference of European Association of Geoscientists andEngineers, Leipzig, ed. EAGE, Proceedings, 10-18.

Yaramanci, U., Lange, G., Knödel, K., 1999. Surface NMRwithin a geophysical study of an aquifer at Haldensleben(Germany). Geophysical Prospecting, 47, 923-943.



Manuscript received March 2003;revision accepted October 2003.

Documents

New technologies in groundwater exploration. Surface ......Geologica Acta, Vol.2, Nº2, 2004, 109-120 U. YARAMANCI The new SNMR method in groundwater exploration 111 where is the gyromagnetic